Mix Challenge - Gossip and Discussion

How to do this, that and the other. Share, learn, teach. How did X do that? How can I sound like Y?
mwaudioprod
KVRist
87 posts since 28 Nov, 2012

Post Sun Aug 17, 2014 3:49 am

Hi there! I have a few questions about average and maximum signal strength and how do I measure it correctly. Did I understand it right that my ready-mix may have a average RMS value of -15dBFS and a maximum true peak at -3 dBFS? What is meant by "average" RMS value? Is meant a 300 ms window, or the whole song? I've looked at the statistics from the last competition and I'm not on the values ​​that are specified there. For example the 3ee Mix, in the PDF it has -20.53 dBFS average RMS and -15.18 dBFS maximum RMS and maximum -4.67 dBFS true peak. Which is the important value, the -20.53 dBFS average RMS or the -15.18 dBFS maximum RMS? My REAPER shows me: Channel 1 = -23.75 dB RMS level and -3.72 dB Peak level, Channel 2 = -23.49 dB RMS level and -4.21 dB Peak level. How can it be that different values ​​are measured? Which free software can I measure that correctly? Means it, that I have here further 8.49 dBFS average RMS available to get to -15 dBFS average RMS (-23.49 dBFS average RMS + 8.49 dBFS average RMS = -15 dBFS average RMS)? This also seems to me very loudly when I compare the -15 dBFS average RMS mix with the other. So many questions. I hope someone can explain to me how I come to the correct values​​.

Compyfox
KVRAF
14288 posts since 19 Oct, 2003 from Berlin, Germany

Re: How do I measure average RMS and maximum Peak signal strength correctly?

Post Sun Aug 17, 2014 4:46 am

Surprising post, but I'll try to get all questions sorted.

mwaudioprod wrote:Did I understand it right that my ready-mix may have a average RMS value of -15dBFS and a maximum true peak at -3 dBFS?
Yes it may.

I recommended to work around -18dBFS / 0 VU (average - in German it's the word "Durschnitt"). But if the signal hovers around -15dBFS (RMS, read again: average) but doesn't exceed -3dBFS, changes are that there were two things happening:

a) some slight pre-mastering already
b) a very dense/compressed sound for the whole production (maybe on purpose, the so called "gluing effect"), which also results in a lower so called "dynamic range"



mwaudioprod wrote:What is meant by "average" RMS value? Is meant a 300 ms window, or the whole song?
If we talk about RMS, VU's or "average measurement time", we always talk about a time frame.
Unless we mention the ITU-R BS.1770 specs (EBU R-128 or ATSC/85 for example), then we need to further specify my mentioning the used meter (i.e. SLk for Short Term Loudness, k-weighted).


The rule of thumb:

An RMS meter uses 300ms as "measurement time frame", which is also known as "integration time" or "ballistics". RMS meters are wrongly called "RMS meters" if you use them in real time - because these meters work best in offline analysis mode to provide a mathematical "average value" (Integrated!) of the whole stream that was analyzed.

Basically speaking during realtime measurements, a VU with no weighting filter, or an RMS meter with no weighting filter, is pretty much the same thing. Only that the look is different, and that the VU shows a shifted scale (i.e. reference level = 0 VU point) while the RMS meter shows full scale values (i.e. -18dBFS average, -15dBFS max).

But the correct usage of a RMS meter is during offline (non real time) statistics analysis (see the following paragraphs).


mwaudioprod wrote:I've looked at the statistics from the last competition and I'm not on the values ​​that are specified there. For example the 3ee Mix, in the PDF it has -20.53 dBFS average RMS and -15.18 dBFS maximum RMS and maximum -4.67 dBFS true peak. Which is the important value, the -20.53 dBFS average RMS or the -15.18 dBFS maximum RMS?

My REAPER shows me: Channel 1 = -20.55 dB RMS level and -3.36 dB Peak level, Channel 2 = -20.48 dB RMS level and -3.18 dB Peak level. How can it be that different values ​​are measured?
I think this is simpler than you might actually think.

If you take note of the statistics sheet, you see three values:
- Average signal strength (RMS, integrated signal stream)
- Maximum signal strength (RMS)
- Loudness Range (ITU-R BS.1770-x)

The "average" (or avg. in short) value is the "mean value" of the measurement. Again, in this case offline meters work best for the mathematics compared to realtime meters. RMS, or "Root Mean Square" meters are basically that - they analyze the whole stream (read: integrated, which would be the correct description), and drop a value of the "average / mean" signal strength, while using 300ms as measurement time frame ever so often.

So much for technical descriptions - very simplified.



What does that mean for you?

Well... if you either used a VU, or a RMS meter in realtime (see above - basically the same thing, but different scale really - unless we talk "offline" with more precise mathematical readouts), then your focus should be on the "average" signal strength and the "max digital peak".


Take a look at my track for example in the statistics sheet:
I mainly used a reference level of -18dBFS / 0VU, but I let my VU go over to about +1dB to +2dB (VU) on forte passages. Forte fortissimo passage overshots (short bursts) could then have resulted in a higher value. The majority of the song (mezzoforte passages) was hovering around the -20dBFS mark. So the "forte" parts of the song was actually around +/- -18dBFS/0VU to -17,5dBFS/+1,5VU during the mixing process.

This gave me the following readouts:
Overall/average signal strength -20dBFS (mathematical, offline => integrated)
Maximal average signal strength -15dBFS
Maximum digital peak (True Peak) -3,5dBTP

I used the system as efficient and as accurate as possible, without bending/breaking the rules too much.



mwaudioprod wrote:Which free software can I measure that correctly?
Pretty much every software that shows max peak.

While using it in realtime, the bargraph or numeric readout in between is your "average" signal strength. Offline measurements can be lower in value (which is still correct).


Note to self:
Actually... thinking about it, maybe Wavelab's statistics doesn't utilize the +3dB AES-17 compensation?! I need to get back in touch with them about this. So thanks for bringing that up!

EDIT:
Just checked, everything seems to be correct.


mwaudioprod wrote:Means it, that I have here further 5.48 dBFS average RMS available to get to -15 dBFS average RMS (-20.48 dB average RMS + 5.48 dB average RMS = -15 dB average RMS)? This also seems to me very loudly when I compare the -15 dB average RMS mix with the other.
Again, you're maybe thinking a bit too complicated.

A peak of up to -15dBFS RMS doesn't mean that the whole production is at that signal strength. Only for specific passages. Which a RMS meter value doesn't show (unless you have access to a histogram).


As example:
Sting - English Man in NY (i.e. here: http://youtu.be/eR3PxNOBDPg )
There is a bridge at the 2:12min mark which then jumps to an aggressive breakdown at the 2:31min mark.

Now, would the majority of the track be like that that passage, the readouts on the RMS meters would be higher. Why? First and foremost, the track doesn't feature that many low end frequencies except for the drum breakdown part (yes, Jazz still has a driving bass, but the majority is mid to high frequencies). So the VU/RMS meter doesn't respond as strong to the bass intensive material - there you go, instant lower readouts.

Furthermore this passage is only like 10-15s longs. But the majority of the track is fairly laid back. This particular passage works as emotional impact. And perceived/felt, it is less troublesome than having the whole song at that signal strength.



Now... how can you objectively compare these productions?

By plain numeric values on a statistics sheet (with RMS values mostly), you can't.
By ear, you can but it takes a lot of time and your ears might tire fairly quick.
By mathematics and the right measurement tool - you can. And this is where the ITU-R BS.1770 specs come into play, and the "loudness normalization" that I did for MC02. Here, different rules apply in terms of measurement time frames and weighting filters (pre-measurement filter curves). But this is a whole world in itself.



Important for you is still:
- what is my average/common/mean signal strength while mixing (overall, the whole track!)?
- do I not overshoot a certain maximum average signal strength?
- what is my maximum digital peak?

Everything else is of matter for the mastering engineer.


mwaudioprod wrote:So many questions. I hope someone can explain to me how I come to the correct values​​.
Lengthy explanation, but hopefully it wasn't confusing. :tu:



EDIT:
And I think this post also makes apparent, why there is a need to tag the numeric values with the correct metering type. i.e. -20dB RMS Integrated, -15dB RMS Max or +3VU max, -4dBTP or -4dBFS digital peak, +10dB DR (Dynamic Range), etc.
[ Studio Page / Twitter ] | [ KVRmarks (see: metering tools) ] | [ Mix Challenge ]

mwaudioprod
KVRist
87 posts since 28 Nov, 2012

Re: KVR Mix Challenge - Gossip and Discussion

Post Sun Aug 17, 2014 6:54 am

Hello Compyfox, thanks for the detailed answering my questions. :tu: I have in my previous post corrected a few numbers. But that should not be decisive. OK, I now understand this: If I use an RMS meter in real time, I should observe a signal strength of -18 dBFS RMS (with 300 ms time frame) and -3 dBFS True Peak in the Master Bus. A signal strength of -15 dBFS RMS and -3dBFS True Peak in the Master Bus results in a slight pre-mastering. Tolerances are common in real-time mixing. The correct RMS values ​​I can only analyze offline (not in real time). Here, the average RMS value (with 300 ms time frame) for the whole song is analyzed (main average). Unfortunately REAPER analyzed other values​​. It is for me still a bit new and confusing, but I'm just learning. By the way, I am also German. :)
EDIT:
And I think this post also makes apparent, why there is a need to tag the numeric values with the correct metering type. i.e. -20dB RMS Integrated, -15dB RMS Max or +3VU max, -4dBTP or -4dBFS digital peak, +10dB DR (Dynamic Range), etc.
I agree! :tu:

Again thank you for your education. Best Regards

Compyfox
KVRAF
14288 posts since 19 Oct, 2003 from Berlin, Germany

Re: KVR Mix Challenge - Gossip and Discussion

Post Sun Aug 17, 2014 4:58 pm

mwaudioprod wrote:Hello Compyfox, thanks for the detailed answering my questions. :tu:
That's why I'm here for. :tu:


mwaudioprod wrote:OK, I now understand this: If I use an RMS meter in real time, I should observe a signal strength of -18 dBFS RMS (with 300 ms time frame) and -3 dBFS True Peak in the Master Bus.
This is what you should shoot for, correct.

Either -18dBFS with a VU with "full scale" numeric scale (volle numerische Skala), or 0VU with a VU that uses a reference level (shifted numeric scale). Or a realtime RMS meter, if it uses 300ms as measurement time frame. Same tool, different look.

mwaudioprod wrote:A signal strength of -15 dBFS RMS and -3dBFS True Peak in the Master Bus results in a slight pre-mastering.
Not necessarily.

-15dBFS RMS while still barely scratching -3dBTP can mean several things:

a) you overdrive the master bus on purpose (i.e. with a saturation module on there to get more coloring)
b) you have more bass intensive signals in the production
c) you mixed hotter than -18dBFS (RMS) / 0VU
d) you have a more dense mix (due to a compressor on the summing bus, the old "SSL Glue" trick comes to mind)

Doesn't mean that you "pre-mastered" anything.


mwaudioprod wrote:Tolerances are common in real-time mixing. The correct RMS values ​​I can only analyze offline (not in real time). Here, the average RMS value (with 300 ms time frame) for the whole song is analyzed (main average).
Tolerances are a huge issue, even at days like this.

But else... yes you summed up the topic fairly well:
RMS (mean) for the whole stream (integrated) can only be mathematically declared in offline mode.


mwaudioprod wrote:Unfortunately REAPER analyzed other values​​. It is for me still a bit new and confusing, but I'm just learning.
The times i used Reaper, I had the impression that they also use 300ms as "ballistics" for the "RMS meter" (bargraph on the master bus). Never dived into the actual statistics to be honest. I'm just used to Wavelab and the third party tools I'm using.

mwaudioprod wrote:By the way, I am also German. :)
Hah! I should start a hotline. ;)


mwaudioprod wrote:Again thank you for your education. Best Regards
Again, don't mention it.
I only hope other KVRians could learn from the post as well.
[ Studio Page / Twitter ] | [ KVRmarks (see: metering tools) ] | [ Mix Challenge ]

User avatar
H-man
KVRian
638 posts since 26 Dec, 2007 from Australia

Re: KVR Mix Challenge - Gossip and Discussion

Post Sun Aug 17, 2014 5:27 pm

Compyfox wrote: I only hope other KVRians could learn from the post as well.
Interesting, and well explained. I will try to "digest" this information. Thx Compyfox.

User avatar
Grant S
KVRer
24 posts since 15 Aug, 2014 from Steubenville, Ohio

Re: KVR Mix Challenge - Gossip and Discussion

Post Sun Aug 17, 2014 6:04 pm

..
Last edited by Grant S on Fri Aug 22, 2014 12:03 pm, edited 1 time in total.

Compyfox
KVRAF
14288 posts since 19 Oct, 2003 from Berlin, Germany

Re: KVR Mix Challenge - Gossip and Discussion

Post Sun Aug 17, 2014 6:17 pm

WOHA!

You're using the EBU R-128 meter already for measuring the statistics?
By these stats alone, you're really running hot.

-15LUFS Integrated and -9,4LUFS SLk max while still only reaching -3,5dBTP max peak...
This tells me that you either premastered the song, or you compressed the production to it's absolute limits.

I'd definitely pull back in terms of loudness in this case. This is a mix challenge (still), not a mastering challenge.


Do you have VU Meter or (realtime) RMS meter at hand?
Or can you do some offline statistics with your host?

Maybe you should take a dive over into the rule book as well:
Mix Participants Guidelines
FAQ (Question 3)


Summed up:
The average and maximum signal strength shall ideally not exceed the following values:

Individual channels:

Average signal strength: -18dBFS (RMS) or 0VU (reference @-18dBFS)
Maximum signal strength: -6dBFS digital peak


Summing bus:

Average signal strength: -15dBFS (RMS) or +3VU (reverence @-18dBFS)
Maximum signal strength: -3dBFS digital peak.
-15LUFS ILk (EBU R-128/ITU-R BS.1770) is not -15dBFS (RMS realtime)/+3 VU (reference @-18dBFS)


I'd say... go back to the "basics" in terms of metering tools again. :tu:



If you're interested to do mastering with the EBU R-128 meter (outside of the KVR Mix Challenge), take a dive into my KVR marks (hint: K-System v2).
[ Studio Page / Twitter ] | [ KVRmarks (see: metering tools) ] | [ Mix Challenge ]

User avatar
Grant S
KVRer
24 posts since 15 Aug, 2014 from Steubenville, Ohio

Re: KVR Mix Challenge - Gossip and Discussion

Post Sun Aug 17, 2014 8:08 pm

,,
Last edited by Grant S on Fri Aug 22, 2014 12:03 pm, edited 2 times in total.

Compyfox
KVRAF
14288 posts since 19 Oct, 2003 from Berlin, Germany

Re: KVR Mix Challenge - Gossip and Discussion

Post Mon Aug 18, 2014 2:33 am

Won't listen to the audio files until the challenge is over.

But one thing for the future:
Ignore the EBU R-128 meter. That one is not suitable during recording/mixing!


Rather the following should apply:
Recording/Mixing: Digital Peak Meter and VU Meter (with reference level) in combination
Mastering: ITU-R BS.1770-x type meter


Everything else would drift this thread too much into a plain metering discussion - which is definitely not my intention in this thread.
[ Studio Page / Twitter ] | [ KVRmarks (see: metering tools) ] | [ Mix Challenge ]

Compyfox
KVRAF
14288 posts since 19 Oct, 2003 from Berlin, Germany

Re: KVR Mix Challenge - Gossip and Discussion

Post Tue Aug 26, 2014 6:18 pm

:!: :!: :!: PUBLIC SERVICE ANNOUNCEMENT :!: :!: :!:
As of MC04, we'll change certain rules of the the "rule set" (read: the guidelines)


The most important stuff summed up:
  • the Mix Challenge is now held bi-monthly (starting September 2014, then in November, January, March, etc)
  • new challenge time frame: 21 days mixing, 3-5 days voting preparations, 14 days voting
  • drop of the participants/4 rule for license winners for the time being - only the winners podium (top three) can win licenses
  • drop of the random voter license winner due to lack of participation
  • it is now mandatory to provide a WAV file for the Loudness Normalization process. An additional MP3 submission is optional
  • Mix Challenge threads are now "multi-purpose" (as with the One Synth Challenge), meaning they are for submissions, voting and feedback
MC03 still utilizes the "participants/4 = license winners" and "random voter license" rule!
The "Guidelines" were just updated.



:arrow: FURTHERMORE

WE NEED MORE SPONSORS!
If you're a programmer, or know companies that offer great plugins and are interested to sponsor the Mix Challenge, please get in touch with either Uncle E or satYatunes.


WE NEED MORE SONGS FOR THE MIX CHALLENGE QUEUE
As of this moment, we have 2 songs in the queue, with great chances to get a third one. But we'd love to see more. Our hearts do beat really high for tracks with vocals, though instrumentals are fine as well. Genre doesn't matter. But interesting would be: Trip Hop, Hip Hop, DnB, Orchestra (with or without Vocals), Acapella/Beatbox, Rock, Heavy Metal. Currently we mainly have access to electronic type tracks.

Please get in touch with the headmaster.


SPREAD THE WORD!
MC03 was backed through News Entries and Twitter comments. A thank you goes out to the Bedroom Producer Blog, the JRR Shop Newsletter and those people that retweeted my Twitter entries. But please, spread the word. The more people join, the more fun. The more sponsors, the more likely we might adjust the rules again so that more people might get chance to win stuff.

Also get in touch with the KVR CEO - we'd still love to see a "Challenge Section" here on the boards for the OSC (One Synth Challenge), the SCC (Song Cafe Challenge) and the MC (Mix Challenge). We already have one section for the Developer Challenge - why not all other KVR hosted/related challenges as well?

So yeah... follow us on Twitter, use the hashtag #kvrmixchallenge , share us on other social media sites, newsletters, forward the challenges to podcasts and webradios. Spread the word! As long as you don't dump a bucket of ice over your head.



[/end of line]
[ Studio Page / Twitter ] | [ KVRmarks (see: metering tools) ] | [ Mix Challenge ]

Compyfox
KVRAF
14288 posts since 19 Oct, 2003 from Berlin, Germany

Re: KVR Mix Challenge - Gossip and Discussion

Post Sat Aug 30, 2014 3:17 am

Teaser for MC04:
The next track will consist of 26 tracks, with a length of 3:27 and a speed of 225bpm (half time 112,5bpm)
It will feature both male and female vocals and not so usual suspects in terms of instruments.


We're currently in preparations for MC04 to start on 1st September (+/- 2-3 days).
Check back every day for a new hint what the next track might be.

Compyfox wrote:We now have a Twitter trend that you can find and follow:
#kvrmixchallenge
Use this hastag and share your thoughts on Twitter, hook up your friends, vote for KVR MC03 (if you have a KVR account and are a bystander), join future Mix Challenges.

Spread the word!
[ Studio Page / Twitter ] | [ KVRmarks (see: metering tools) ] | [ Mix Challenge ]

Compyfox
KVRAF
14288 posts since 19 Oct, 2003 from Berlin, Germany

Re: KVR Mix Challenge - Gossip and Discussion

Post Sun Aug 31, 2014 4:14 pm

Teaser for MC04:
The next track will definitely rattle things up a bit with some "Ska Punk".


We're currently in preparations for MC04 to start on 1st September (+/- 2-3 days).
Check back every day for a new hint what the next track might be.

Compyfox wrote:We now have a Twitter trend that you can find and follow:
#kvrmixchallenge
Use this hastag and share your thoughts on Twitter, hook up your friends, vote for KVR MC03 (if you have a KVR account and are a bystander), join future Mix Challenges.

Spread the word!
[ Studio Page / Twitter ] | [ KVRmarks (see: metering tools) ] | [ Mix Challenge ]

camsr
KVRAF
6931 posts since 17 Feb, 2005

Re: KVR Mix Challenge - Gossip and Discussion

Post Sun Aug 31, 2014 9:31 pm

Was their going to be a mastering challenge also?
Image

bjporter
KVRAF
2605 posts since 19 Dec, 2010 from North America

Re: KVR Mix Challenge - Gossip and Discussion

Post Sun Aug 31, 2014 11:55 pm

Meesa lookin forward to theese :hyper:

Compyfox
KVRAF
14288 posts since 19 Oct, 2003 from Berlin, Germany

Re: KVR Mix Challenge - Gossip and Discussion

Post Mon Sep 08, 2014 3:43 am

Since nobody answered this yet:
camsr wrote:Was their going to be a mastering challenge also?
We have something along those lines in the making, but can't give any specific details yet.





:!: :!: :!: PUBLIC SERVICE ANNOUNCEMENT :!: :!: :!:
I just updated the "Guidelines" once more, to make it even more clear, that the "Mix Challenge" has a focus on creating a balanced mixdown (and mixdown only!), not creating a pre-mastered/release ready end product.

Mixing these days sadly has an obscure reputation, that a track has to be (pre)mastered as well in order to be heard. Whoever made that up years ago, talked nonsense IMO. And this is not what this Mix Challenge is about.
[ Studio Page / Twitter ] | [ KVRmarks (see: metering tools) ] | [ Mix Challenge ]

Return to “Production Techniques”